Canvas: Discussion Summary

What is Discussion Summary?

Discussion Summary is an AI-powered course feature in Canvas. It creates a custom summary of posts in a Canvas Discussion. The tool uses an AI model to detect patterns in the most recent discussion contributions. You can customize the summary prompts to guide the output. The summary is only visible to the instructor.

How to turn on Discussion Summary in a course?

Discussion Summary is off by default. Turn it on at the course level:

1. Click Settings in the course navigation menu in a course.

2. Click the Feature Options tab.

3. Find Discussion Summary and lick the red X under State to enable it. When enabled the icon will present as a green check mark.

How does Discussion Summary work in Canvas?

When enabled, people with the Teacher role can access a "Summarize" button in a Discussion activity. When prompted, data for the full prompt and all replies (including any personally identifiable information) are transmitted to the AI model for single use processing. When summary data is returned, Canvas saves the text in its database, but no data stays with the model. Instructors can edit or regenerate summary outputs before sharing.

What are some best practices for instructors?

1. Treat any AI output from the Discussion Summary tool like a draft, not a final product. Large Language Model (LLM) outputs are not always accurate or reliable.

2. Keep prompts focused. Clear, specific discussion questions yield clearer summaries. In testing, the AI summary output was often overly general to be helpful on its own. Test alternative prompts that could return output with more specificity. Here are some prompt suggestions to try:

   • Summarize the main themes related to [topic] and their frequency.

   • What concepts generated student confusion or misconceptions?

   • What questions did students most commonly ask?

   • Summarize points of disagreement or differing views.

   • Highlight examples of student interaction (replies, building on ideas).

   • Identify potential gaps in the discussion coverage of [topic].

   • What information might have been helpful before the discussion?

   • Note connections made to real-world examples or other course material.

   • Describe the types of evidence or support used by students.

3. Check for personally identifiable information (PII) and remove or anonymize personal details that may have surfaced in the output before sharing.

4. Turn the feature off if it is not meeting instructional needs. Disable the feature in Settings  > Feature Options per course.

What are Discussion Summary data privacy considerations?

• The AI model is not trained on user data. The model is pre‑trained. This means it does not learn from prompts or discussion post data.

• Names or personal details shared in discussion posts are visible to the model during processing.

• Summary output data is stored in the Canvas database. This means Instructure does not send them back to the model.

How is embedded media handled in summary output?

Images, video, and audio posts are ignored. Alt‑text or transcripts may be included in future updates.

Can a summary be regenerated? 

Yes. click Regenerate and review the new draft.

Will students see my edits? 

Only after you click Post Summary. Until then, the text is private to you.

Multilingual support? 

The model handles many languages, but accuracy may vary.

How to get additional help?

For more information visit Canvas’ Nutrition Facts for Discussion Summaries page.

For technical support, contact Instructure Support by 24/7 phone [+1-833-890-4166] or email support@instructure.com. For pedagogical support schedule a consultation with a DLINQ team member.